Predictor Elimination in Image Coding Reducing Bitrate

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image processing methods, such as those in the H264/AVC standard, face inefficiencies due to the large number of predictors used in predictive coding, leading to increased bitrate and high processing power consumption, especially when certain codings are similar, resulting in unnecessary information transmission and excessive processing demands.

Innovation Solution

A method that reduces the number of predictors by eliminating redundant ones through a predictor elimination algorithm, which calculates differences between predictors and applies a quantization matrix to determine optimal elimination, allowing for content-adaptive coding that is reproducible at both the encoder and decoder levels.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If multiple predictors are used in predictive coding, then coding accuracy is improved, but bitrate increases due to competition information

Engineering Contradiction:
Improvecoding accuracyVSAvoidbitrate
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

The patent extracts and eliminates redundant predictors from the set of competition predictors. By analyzing the redundancy among predictors and removing those that provide minimal additional information, the system maintains coding accuracy while reducing the bitrate required to transmit competition information.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent dynamically changes the number of predictors based on image content characteristics. By adapting the predictor set size according to the specific image being encoded, the system optimizes the balance between coding accuracy and bitrate efficiency for different content types.

Inventive Principle:
Principle #35Parameter changes

2Measurement precision

If multiple predictors are used in predictive coding, then coding accuracy is improved, but processing power consumption increases

Engineering Contradiction:
Improvecoding accuracyVSAvoidprocessing power
Core Design Contradiction:
Measurement precisionVSPower

Solution Approach 1:

The patent extracts and eliminates redundant predictors to reduce the computational workload. By removing predictors that contribute minimally to coding accuracy, the system significantly reduces the processing power required for predictor evaluation and competition while maintaining coding performance.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent applies partial action by using only the necessary number of predictors rather than all possible predictors. This approach performs sufficient prediction operations to achieve good coding accuracy without the excessive computational burden of evaluating all potential predictors.

Inventive Principle:
Principle #16Partial or excessive action

3Manufacturing precision

If all predictors are evaluated with selection criterion, then coding optimization is improved, but calculation complexity increases

Engineering Contradiction:
Improvecoding optimizationVSAvoidcalculation complexity
Core Design Contradiction:
Manufacturing precisionVSDevice complexity

Solution Approach 1:

The patent extracts and removes redundant predictors before the selection criterion evaluation stage. By reducing the number of predictors that need to be evaluated, the system maintains coding optimization capability while significantly lowering calculation complexity and CPU processing requirements.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent performs preliminary elimination of redundant predictors before the main selection criterion calculation. This preliminary action reduces the scope of subsequent optimization calculations, making the overall process more efficient while still achieving good coding optimization results.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP2351373B1Encoding and decoding with elimination of one or more predetermined predictors
Publication Date: 2018.07.04 ORANGE SA
  • EP2351373B1 patent drawingFigure 1
  • EP2351373B1 patent drawingFigure 2~4B
  • EP2351373B1 patent drawingFigure 5

AI summary

The invention relates to a method for encoding an image or a sequence of images generating a data flow (F) comprising data representing at least one group of pixels (MB) in one of said images (IE), said method including the steps of: calculating a predetermined number of predictors (P1,..., Pn) in relation to said group of pixels (MB), and selecting an optimal predictor in accordance with a predetermined selection criterion. According to the invention, between the calculation and selection steps, a step of eliminating at least one predictor with a calculated predetermined number is performed, which consists, for at least two ith (Pi) and jth (Pj) predictors, of: calculating the difference between said jth and ith predictors, applying a transform to said calculated difference, performing a quantification operation on the result of the applied transform, comparing the result of said quantification operation with a predetermined value, eliminating the jth predictor if the result of said quantification operation is less than or equal to said predetermined value.